"Again, the basic economics of housing prices are pretty simple—supply and demand. ... The actual marginal cost of adding an extra square foot of living space at the top of a skyscraper in New York is typically less than $400. Prices do rise substantially in ultra-tall buildings—say, over 50 stories—but for ordinary skyscrapers, it doesn’t cost more than $500,000 to put up a nice 1,200-square-foot apartment. The land costs something, but in a 40-story building with one 1,200-square-foot unit per floor, each unit is using only 30 square feet of Manhattan—less than a thousandth of an acre. At those heights, the land costs become pretty small. If there were no restrictions on new construction, then prices would eventually come down to somewhere near construction costs, about $500,000 for a new apartment. ... In a sample of condominium buildings, I found that more than 80 percent of Manhattan’s residential buildings built in the 1970s had more than 20 stories. But less than 40 percent of the buildings put up in the 1990s were that tall."All this makes me curious as to just how much of New York's residential space is actually located in tall buildings. After all, if skyscrapers are to be the remedy for New York's high housing costs, they must be capable of providing a substantial fraction of the city's housing supply.
Fortunately, the Council on Tall Buildings and Urban Habitat has published a list of all New York buildings 20 stories and taller (the arbitrary cut-off used by the Council, but mentioned by Glaeser as well). In New York City, including all the five boroughs, there are 552 such buildings completed as of 2010. Of these, only 169 are devoted exclusively to residential use, with an additional 26 being partly residential. The average height for these buildings is approximately 41 stories.
In all of New York, there are a total of 932,706 residential parcels, according to the New York Property Tax Report for 2010. That is, buildings 20 stories and more with at least some residential component comprise two-hundredths of one percent of all taxable residential parcels in New York City. Granted, very tall buildings are likely to have far more units than the average residential parcel — examples include the 40-story Chelsea Stratus with 235 units, the 41-story One Sutton Place North with 334, the 43-story Platinum with 185, the 43-story America Apartments with 211 and the 42-story 325 5th Avenue with 250.
The same tax report puts the number of individual residential units in New York at 2.93 million (this of course omits a very large amount of New York-accessible housing in New Jersey, Connecticut, and Westchester and Nassau Counties, very little of it in tall buildings). Assuming the average residential tower to have 250 units, and factoring in the 26 part-residential buildings, we get (169*250) + (26*125) = 45,500/2.93 million = 1.6%. That is, units located in "tall" buildings comprise 1.6% of the residential units in the City. But wait: if the claim is that tall buildings are needed to supply additional housing, not the absolute number of units should be counted, but only those provided by the "tall" portion of the building — that portion 20 stories and over. Reducing the figures accordingly, only .80% of all units in the city are located above a 19th floor level, or, 1 out of every 125 units. And even this is probably a substantial overestimate, as it is likely that proportionately more units are located on lower stories than higher ones (due to a narrowing of towers as they rise, in combination with larger floorplans being located on higher floors — see below).
A few observations on all this: first, it's clear that skyscrapers' visual impact is all out of proportion to their real estate impact. Even an unprecedented residential skyscraper building boom, such as the one during the past decade, constitutes a very small portion of the overall increase in New York's housing supply. The great bulk of the increase evidently derives from low and mid-rise construction. Second, the seemingly low number of units in several of the high rises appears to be a result of their ultra-luxurious nature: the expense of building very high necessitates very high prices, which in turn requires ample floorplans on desirable higher floors to attract high-income buyers. I suspect this produces a declining return on the total number of units per floor as one builds higher (other reasons for internal space constraints on high-rise construction can be found here). Finally, in the cases of many of these towers, but particularly those from the 1960s and 70s, the buildings do not make efficient use of their lot (for various and debatable reasons). In some cases, lot coverage appears to be as low as 30 or 40 percent. Were the entire lot to be used, the same number of units could be accommodated at one-half the height or less.
As to Glaeser's observation about the percentages of towers built in the 70 and 90s: the actual numbers for those are 20 towers built in the 1970s and 21 in the 1990s. Unmentioned are the 2000s, in which 56 such towers were built in New York. Why the 90s were used by Glaeser for comparison instead of the 2000s, I can only guess.